Collin Allen's questions

In the Red Hat Software Collections MySQL container, I see that you can set MYSQL_DEFAULTS_FILE to indicate the path to a completely different my.cnf file. When used in OpenShift, I don't understand how to get such a config file on disk, since their S2I scripts have already been run as part of the mysql-persistent template. How do I go about injecting such a file into a pre-built image?

I'm able to deploy MySQL 5.7 using the template provided with OpenShift, and can set other environment variables, but am not sure how to put my own files into the image so I can reference them.

While configuring Varnish 4, I'm interested in short-circuiting inside vcl_recv, but I don't want to skip the builtin.vcl (née default.vcl) VCL logic. For example, given this VCL pseudocode:

sub vcl_recv {
    if (somecondition1) {
        set some_thing;
        return (hash); # Return from cache if present, or fetch from the backend
    }

    if (somecondition2) {
        set some_other_thing;
        return (hash); # Return from cache if present, or fetch from the backend
    }

    // ...

    return (pass); # Skip cache, fetch from the backend
}

Question 1

If I'm calling return in every possible code path, then the built-in logic will get skipped, right?

Question 2

Is there a way to format code similar to the above, but without skipping the builtin.vcl logic? The only way I can see is to combine my conditionals, boolean ! them, and move the return (pass); inside the giant if.

I'm attempting to install psycopg2 into a Python virtualenv with Ansible's pip module, but I need to prepend an entry to PATH for it to build correctly (it needs to know the path to the directory containing pg_config). I see that I can pass environment to the pip module, but I'm unsure of how to prepend rather than overwrite PATH.

Here I'm attempting to prepend the path with the necessary directory, but it overwrites the virtualenv PATH and fails:

- pip:
    name: psycopg2
    virtualenv: /path/to/my/venv
  environment:
    PATH: /usr/pgsql-9.3/bin:$PATH

I've got an nginx config for a PHP web app and I would like to make it serve an additional purpose, and I can't seem to find a way to accomplish it. My current nginx config does the following:

• Rewrites / and /intro/* to static HTML pages for a non-PHP powered "marketing" site
• Attempts to load /maintenance.html if present on disk, which only exists during deploys
• If /maintenance.html does not exist, sends requests to PHP via FastCGI

The end result is that the static pages remain available during deploys, and the back end PHP app can go down for maintenance while deploys happen. The important portion of the server config block looks like:

rewrite ^/$ /marketing/index.html last;
rewrite ^/intro/(.*)$ /marketing/$1.html last;

try_files $uri /maintenance.html @webapp;
location @webapp
{
    rewrite ^(.*)$ /index.php?url=$1 last;
}

include php.conf;

(php.conf config handles location ~ \.php$ locations and does all the usual fastcgi_param setting. Nothing project-specific there, so I didn't post that block.)

What I would like to be able to do in addition to the above is when "maintenance mode" is on, serve up /maintenance.html for standard HTTP requests, but respond with some JSON when handling AJAX requests. I've found that I can safely return rudimentary content by doing:

if ($http_x_requested_with = XMLHttpRequest)
{
    return 503 '{"maintenance":true}';
}

However, this obviously happens for all requests, not just those where /maintenance.html is being served. It seems I will need some flag somewhere, since try_files doesn't let you do additional things depending on which file or location it selected.

Is there some way nginx can be configured to return both HTML and JSON maintenance mode details?

Edit: Thinking about it more, it would be nice to perhaps move maintenance.html out of try_files completely, so that it can be served with a 503 status instead of a 200.

I've got a test instance of a compiled web app (that I didn't write) which I'd like to do some testing with, however the app contains client email addresses that I would like to not receive emails while I do said testing.

Right now, this web app uses a particular SMTP server — call it mail1 — to send out notification emails when various events happen. To minimize the amount of customization I would need to do to the app or its data (which may skew the testing), I'm thinking a solution to avoiding sending these emails might be to stand up a temporary SMTP relay mail2 that inspects the outbound email addresses and only relays emails to mail if the emails are destined for a particular domain. All others not destined for that domain can be dropped, or even better, forwarded to a different address on that whitelisted domain to see that they would have been sent to clients if not for the whitelisting in mail2.

Is this the sort of thing I could do with Postfix or Qmail? If so, are there any recommendations for how to get started doing this? I've never configured either before, and I'm wondering if there's perhaps an entirely different mail-limiting solution that I'm overlooking. Maybe even something Gmail/Google Apps can do?

My company has grown to the point where we're interested in being an intermediate CA for our customers, and would like to issue certificates to customers. Among other uses, we wish to issue SSL certificates that are trusted by the average web surfer's browser.

My question is quite similar to this one, but I'm skeptical of the "You don't" answer. You must be able to, given the time and money -- others do it. We have both, and would like to know how to proceed.

Under Exchange 2007, can a single Active Directory user have multiple unique mailboxes? When attempting to create a second mailbox for a given user, that user is no longer in the list of choose-able ones, since they already have a first mailbox.

I'm aware that a single mailbox can have multiple SMTP addresses associated with it, but the downfall I'm running into is when creating a new Distribution Group. It seems you can only associate users with a Distribution Group, but I want to send to a particular email address, which is often not the default reply for that user.

In our current non-Exchange IMAP setup, each email address is its own mailbox, and some users have several. Aliases (distribution groups, effectively) become straightforward, as messages can be delivered to a specific set of email addresses.

Am I doing something wrong, or just approaching this the wrong way? I could make a new Active Directory account & Mailbox combo for each address I wish to catch, but that seems like the wrong thing to do. I'm really new to Exchange, so any hints are much appreciated.

For my particular setup, I've got control of a BIND 9 zone file for, say, example.com. There are several hosts in this zone, such as www.example.com and mail.example.com. I want the public internet to be able to query these hosts and retrieve their IPs as you'd expect.

Can some hosts be flagged as "private", in the sense that most public internet users can't query information about these hosts, except for a specific IP range of users? For example, can I make a third host, secret.example.com query-able only by users in the 192.168.0.x range?

I'm aware that I can do this with separate zones in BIND, but this doesn't seem to offer what I need. The key here is that both the public and private hosts need to be part of the same parent, example.com in this case. This could also be achieved by the /etc/hosts file on the few privy machines, but then the records would not be centrally manageable.

Is this possible, or am I overlooking a different solution?